[darcs-users] Darcs memory usage

[Jorey Bump]

Wolfgang Dobler wrote:
> Juliusz Chroboczek writes:
>  > > I have been using Darcs to version-control the ASCII dumps of a Postgresql
>  > > database. The dump is ~ 15 MB in size.
>  > 
>  > Darcs is the wrong tool for this job.  For keeping machine-generated
>  > files of that size, I would recommend a snapshot-based rather than
>  > delta-based tool.
> 
> I just thought the deltas weren't that bad for my case, as the dump
> process produces similar files each time.

I also use darcs in this fashion, but for smaller databases, which may 
be the reason I'm not experiencing the same problem. I've become 
completely dependent on darcs for version controlling just about 
everything, including code, documentation, web sites, system 
configuration, etc.

But for point-in-time backups, I've been using rdiff-backup:

  http://www.nongnu.org/rdiff-backup/

If you like darcs, as I do, you might love rdiff-backup. Based on the 
rsync protocol, it provides very efficient incremental backups, and can 
even store them on remote hosts. Restoring or retrieving copies of files 
or directories from any point-in-time is a snap.

During the day, I version control. Once a night, rdiff-backup takes 
snapshots (including darcs repositories). This is pretty good insurance 
against most mishaps. If you're only interested in saving efficient 
snapshots of your data (without the event-driven emphasis of version 
control), consider dumping your data regularly and making backups with 
rdiff-backup.

I'm currently working on a project that contains ~ 500 MB of files. As 
much as I'd like to version control it with darcs, I don't want to 
devote over 2 GB of storage to a single project (working + pristine + 
patches *and* backup of repository). So I keep a changelog in the 
directory and manually take snapshots with rdiff-backup. It's primitive, 
but it works.